完整後設資料紀錄
DC 欄位語言
dc.contributor.author林佳靜en_US
dc.contributor.authorLin, Jia-Jingen_US
dc.contributor.author陳安斌en_US
dc.contributor.authorChen, An-Pinen_US
dc.date.accessioned2014-12-12T01:44:35Z-
dc.date.available2014-12-12T01:44:35Z-
dc.date.issued2009en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#GT079764501en_US
dc.identifier.urihttp://hdl.handle.net/11536/46232-
dc.description.abstract本研究應用資訊技術的人工智慧分類及預測模型來建立台灣銀行業的信用評等模型。以2004年至2009年為研究實驗區間,國內銀行業為研究實驗對象,選取1979年美國聯邦金融機構檢查評議委員會提出銀行評等制度,以安全性及健全性作為指標基礎,包括:資本適足性、資產品質、管理能力、盈利性及流動性等。進行自組織映射神經網路分類評估及倒傳遞類神經網路特徵學習,期望能降低評等機構之人為主觀性,更快速反映銀行業資產體質揭露。
多年來的財務槓桿化被視為支撐銀行業資產品質的要因。而美國銀行業過度的財務槓桿運作,由各國政府、銀行及投資大眾共同分攤風險。在2007年中∼2008年因連動債發生全球信貸危機與金融風暴,國際信評公司在這波金融風暴,預警制度卻完全失靈。信用評等的公信力明顯失控,導致全球各界開始質疑外部信評機構的即時性、公正性。若能建立資訊技術來降低人為主觀性的信用評等模型來檢定企業體質,將可降低信評機構人為因素的過度干擾評等之公正性,並提升評等的可信度及透明度等問題。
實證研究結果顯示,對於以財務比率指標為變數,可預測銀行業信用評等體質的強弱。本研究亦證實透過自組織映射圖神經網路與倒傳遞類神經網路的結合,預測評等體質強弱之準確度優於倒傳遞類神經網路,另外亦發現將國際信用評等納入考量,確實有較佳的體質強弱預測表現。
zh_TW
dc.description.abstractThis research applies IT’s artificial intelligence classification and forecast model to establish a credit rating model for the banking industry in Taiwan. In this research, we use domestic banking industry as the study object and the study period is between year 2004 and 2009; we choose the credit rating system proposed by the U.S. Federal Financial Institutions Examination Council in 1979, which is based on the level of security and soundness, including capital adequacy, asset quality, management ability, profitability and liquidity. We perform Self-Organization Map for classification and evaluation, as well as Back-Propagation Network for characteristic learning. It is expected to reduce credit rating agency’s subjectivity and to promptly reveal the asset condition in banking industry.
For years, the leveraged finance has been considered as the reason that supports the asset quality of banking industry. The risk from U.S. banking industry’s excessive operation of leveraged finance has been shared by various governments, banks and investors. The structured note caused the global credit and financial crisis in year between 2007 and 2008. The forecasting system of international credit rating company was not working properly during this financial crisis. The public credibility of credit rating was obviously out of control, and this causes people to doubt credit rating agency’s ability of instantaneous awareness and fairness. If we are able to establish a credit rating model with information technology and lower subjectivity to evaluate enterprises’ financial condition, then we can reduce the human factor from credit rating agency that causes the fairness and increase the reliability and visibility of credit rating.
From the experimental results, to evaluate banking industry's credit rating with financial ratio index as variables. This research is based on the combination model of Self-Organization Map and Back-Propagation Network, besides we found that the integrate model is superior to the back-propagation neural network model. In addition, we also discover that the forecasting model works better if the international credit rating is taken into consideration.
en_US
dc.language.isozh_TWen_US
dc.subject信用評等zh_TW
dc.subject自組織映射圖神經網路zh_TW
dc.subject倒傳遞類神經網路zh_TW
dc.subjectCredit Ratingen_US
dc.subjectSelf-Organizing Map Neural Networken_US
dc.subjectBack-Propagation Neural Networken_US
dc.title應用類神經網路於台灣銀行業信用評等模型之研究zh_TW
dc.titleApplying the Artificial Neural Networks on The Study of Bank Credit Rating Model in Taiwanen_US
dc.typeThesisen_US
dc.contributor.department管理學院資訊管理學程zh_TW
顯示於類別:畢業論文